WO2018199434A1

WO2018199434A1 - Method for predicting state of life of battery on basis of numerical simulation data

Info

Publication number: WO2018199434A1
Application number: PCT/KR2018/000917
Authority: WO
Inventors: 김광선; 장경민; 주강우
Original assignee: 한국기술교육대학교 산학협력단
Priority date: 2017-04-25
Filing date: 2018-01-19
Publication date: 2018-11-01
Also published as: US10254349B2; US20180306868A1; KR101792975B1

Abstract

The present invention relates to a method for predicting a state of life of a battery on the basis of numerical simulation data. A method for predicting a state of life of a battery by a battery management system according to an embodiment of the present invention comprises the steps of: when a numerical analysis result is verified using a laboratory test result obtained through electrical and chemical analysis of a battery, acquiring a numerical simulation database obtained by extracting and storing solution data of the battery; counting the number of times of charging or discharging when a difference between reference data read from the verified numerical simulation database and measurement data read from the battery falls within a predetermined range and a battery capacity satisfies a predetermined condition; and predicting a state of life of the battery by using the number of times of charging or discharging and a classifier according to a pre-learned machine learning algorithm.

Description

How to predict the life status of batteries based on numerical simulation data

The present invention relates to a method for predicting the life state of a battery based on numerical simulation data, and more particularly, based on numerical simulation data capable of predicting the life state of a battery based on a numerical simulation database in which numerical analysis results are verified. The present invention relates to a battery life state prediction method.

Recently, people are starting to use lithium-ion batteries for automobiles and Energy Storage System (ESS). At the same time, high energy densities that maintain long operating times are emerging as important research. They solve the problem by inserting the various cells in one pack with proven output and stability.

However, excessively discharging or charging a high energy density battery can negatively affect battery life when its characteristics are not properly identified. Packed multiple cells or even one cell, and heat generated from the environment, can in particular shorten the life of the overall pack performance. Shorter lifespan and initial malfunction of battery packs occur mainly due to unexpected changes in the physical properties of each cell when lithium ion batteries are repeatedly charged and discharged. Because of this important impact of battery material properties on pack performance, there have been various numerical simulation studies of modified properties.

Computer simulations, on the other hand, are increasing due to more reliable engineering numerical analysis methods, along with high-speed data transfer and processing in computer systems.

1 is a diagram showing 50 battery impedances manufactured by the same process. 2 is a view showing 50 battery capacity manufactured by the same process.

Medium and large battery packs for hybrid vehicles and ESSs require high capacity and high power systems with dozens or hundreds of single cells managed by BMS.

1 and 2, lithium ion batteries have different internal characteristics (eg, impedance, capacity, etc.) even if they are manufactured in the same factory in the same process. This means that the batteries you connect have different characteristics and thus create different charge and discharge rates for the pack. The battery pack is managed by the BMS. The BMS stops the charging and discharging rate of the cell, even if other cells still need time to charge or discharge, to prevent inefficient operation and to prevent accidents that may result from overcharging or overdischarging.

3 is a diagram illustrating different charging and discharging voltages of respective batteries.

3 shows that different characteristics of each battery provide different charge and discharge rates for three different batteries. Recent commercially available Li-ion battery packs BMS can monitor temperature, voltage, current and SOC (eg, state of charge or discharge) in real time. The BMS of lithium ion battery packs can control and manage the electrical circuits of battery cells or packs if their monitoring values exceed the normally expected values.

This kind of BMS cannot predict the reliable lifetime of each cell with different characteristics because it is very difficult to obtain predicted data from actual laboratory experiments about other effects such as the battery's natural aging or temperature changes.

Even in some possible limited conditions, including thermally balanced pack designs, various cases of actual laboratory experiments on other Li-ion batteries can be very time consuming and expensive.

Embodiments of the present invention provide a method for predicting the life state of a battery based on numerical simulation data, which can reliably predict the life state (SOH) of the battery based on the numerical simulation data.

Embodiments of the present invention provide numerical simulation data for predicting SOC and SOH based on numerical sample cases with reliable big data predicted with respect to battery capacity (SOC) and battery life state (SOH). The present invention provides a method for predicting the life state of a battery.

Embodiments of the present invention provide a method for predicting the life state of a battery based on numerical simulation data, which can provide an alarm or a replacement signal to a BMS when an abnormal operation continues for a predetermined time within a tolerance.

However, the object of the present invention is not limited to the above-mentioned object, and although not mentioned, other objects according to the following means or specific configurations in the embodiments will be clearly understood by those skilled in the art from the description. will be.

According to a first aspect of the present invention, in the method for predicting the life state of a battery performed by a battery management system, the solution of the battery when the numerical analysis result is verified as a laboratory result through electrical and chemical analysis of the battery Obtaining a numerical simulation database from which data is extracted and stored; Counting the number of charges or discharges when the deviation between the reference data read from the verified numerical simulation database and the measurement data read from the battery is within a preset range and the battery capacity satisfies a preset condition; And estimating the life state of a battery using a classifier according to the number of charges or discharges and a machine learning algorithm that has been learned. The method for predicting a life state of a battery based on numerical simulation data may be provided.

The acquiring step may include setting an initial condition for a battery; Performing numerical analysis through electrical and chemical analysis on the battery; Verifying the performed numerical analysis result as an experimental result; Extracting solution data of a battery for the numerical analysis result when the verification is completed; And storing the extracted solution data in a verified numerical simulation database and obtaining from the verified numerical simulation database.

The performing of the numerical analysis may use at least one of a Species Transport Model, an Electronic Potential Model, a Chemical Reaction Model, and an Energy Balance Model. Numerical analysis.

The counting may include: counting the number of charges when the battery is being charged and the battery capacity is fully charged and stopping charging; counting the number of discharges when the battery is being discharged and the battery capacity is the minimum battery capacity; Can be.

The counting step may include reading reference data from a validated numerical simulation database; Reading measurement data from a battery that is being charged or discharged; Calculating a deviation between the read reference data and the read measurement data; Checking whether the calculated deviation is within a preset range and calculating a battery capacity if the calculated deviation is within a preset range; And counting the number of charges or the number of discharges and stopping the charging or discharging when the calculated battery capacity is full or less than the minimum battery capacity.

The predicting step may be performed using a machine learning algorithm of any one of a support vector machine, Bayes classifiers, artificial neural networks, and decision trees. The state can be predicted.

The life state prediction method may further include emergency stopping the charging or discharging of the battery through a safety algorithm when a deviation between the reference data and the measurement data is out of a predetermined range.

The emergency stopping may include reading an nth error value and an n−1th error value from an error database; calculating a sum of error deviation values of deviations of the n th error value and the n-1 th error value; Checking whether a preset maximum error value is equal to or less than the sum of the calculated error deviation values; Calculating a battery capacity at the time of charging or discharging, when the preset maximum error value is not equal to or less than the sum of the calculated error deviation values; And emergency stopping the charging or discharging operation of the battery when the preset maximum error value is less than or equal to the calculated error value sum.

Meanwhile, according to the second aspect of the present invention, in the method of predicting the life state of a battery performed by a battery management system, when the numerical analysis result is verified as a laboratory result through electrical and chemical analysis of the battery, the battery Obtaining a numerical simulation database of extracted and stored solution data; Compare the measured data read from the battery with the reference data read from the verified numerical simulation database, and the charging or discharging time is within the maximum and minimum allowable time for reaching the target voltage, and the battery capacity satisfies the preset conditions Counting the number of charges or discharges; And estimating the life state of the battery using a classifier according to the counted number of charges or discharges and the previously learned machine learning algorithm.

The counting step may include reading reference data from a validated numerical simulation database; Reading measurement data from a battery that is being charged or discharged; Comparing measured data read from a battery with reference data read from the verified numerical simulation database; Check whether the charge or discharge time is within the maximum and minimum allowable time to reach the target voltage, and if the charge or discharge time is within the maximum and minimum allowable time to reach the target voltage and the battery capacity meets the preset conditions Calculating a dose; And counting the number of charges or the number of discharges and stopping the charging or discharging when the calculated battery capacity is less than or equal to the full charge or the minimum battery capacity.

The life state prediction method may further include emergency stopping the charging or discharging of the battery through the safety algorithm when the charging time or the discharging time is out of the maximum and minimum allowable time for reaching the target voltage.

Embodiments of the present invention can reliably predict the life state (SOH) of a battery based on numerical simulation data.

Embodiments of the present invention may predict SOC and SOH based on numerical sample cases with reliable big data predicted in relation to battery capacity (SOC) and battery life state (SOH).

Embodiments of the present disclosure may provide an alarm or replacement signal to the battery management system when abnormal operation continues for a certain time within a tolerance.

1 is a diagram showing 50 battery impedances manufactured by the same process.

2 is a view showing 50 battery capacity manufactured by the same process.

4 is a diagram illustrating a configuration of a battery management system to which a method for predicting a life state of a battery based on numerical simulation data according to an embodiment of the present invention is applied.

5 is a diagram illustrating a conceptual diagram of a battery to which a one-dimensional battery model and a two-dimensional jelly roll model are applied.

6 is a diagram illustrating a lithium ion battery charging curve at various temperatures.

7 is a diagram illustrating a discharge curve of a lithium ion battery at various temperatures.

8 shows the time (1 cycle) to reach 4V while charging at different temperatures.

Fig. 9 shows the time (1 cycle) before reaching 3.5V while discharging at different temperatures.

10 is a flowchart illustrating a numerical simulation database verified by numerical analysis according to an exemplary embodiment of the present invention.

FIG. 11 is a flowchart illustrating a life state of a battery using a machine learning algorithm according to an exemplary embodiment of the present invention.

12 is a flowchart illustrating an SOH prediction algorithm based on real-time measured voltage during charging according to an embodiment of the present invention.

FIG. 13 is a flowchart illustrating an SOH prediction algorithm based on real-time measured voltage during discharge according to an exemplary embodiment of the present invention.

14 is a flowchart illustrating a safety algorithm for emergency stop according to an exemplary embodiment of the present invention.

15 is a flowchart illustrating an SOH prediction algorithm based on a time for reaching a maximum voltage during charging according to an embodiment of the present invention.

16 is a flowchart illustrating an SOH prediction algorithm based on a time for reaching a maximum voltage during discharge according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. It will be described in detail focusing on the parts necessary to understand the operation and action according to the present invention. In describing the embodiments of the present invention, descriptions of technical contents that are well known in the technical field to which the present invention belongs and are not directly related to the present invention will be omitted. This is to more clearly communicate without obscure the subject matter of the present invention by omitting unnecessary description.

In addition, in describing the components of the present invention, different reference numerals may be given to components having the same name according to the drawings, and the same reference numerals may be given to different drawings. However, even in such a case, it does not mean that the corresponding components have different functions according to embodiments, or does not mean that they have the same functions in different embodiments. Judgment should be made based on the description of each component in.

In addition, the technical terms used in the present specification should be interpreted as meanings generally understood by those skilled in the art to which the present invention belongs, unless specifically defined otherwise in the present specification, and excessively comprehensive It should not be construed in meaning or in excessively reduced sense.

Also, the singular forms used herein include the plural forms unless the context indicates otherwise. In the present application, terms such as “consisting of” or “comprising” should not be construed as necessarily including all of the various components or steps described in the specification, and some of the components or some steps It should be construed that it may not be included or may further include additional components or steps.

As shown in FIG. 4, the battery management system 100 according to an exemplary embodiment of the present invention includes a controller 110, a data input / output unit 120, a numerical simulation database 130, and an error database 140. .

The battery management system 110 (BMS) may include an aging degree (SOH: State of Health, or a life state of a current battery) and a state of charge (SOC: State of Charge, or a current battery) of a battery cell or the battery pack 10. The voltage (V), current (I) and temperature (T) of the battery cell (or battery pack) are monitored for the monitoring and management of the status.

The battery pack 10 is connected to the load 11 and provides the stored energy to the load 11.

A battery management system (BMS) according to an embodiment of the present invention may more accurately predict the life state (SOH, or aging degree) of a battery using reference data from a measurement data and a numerical simulation database measured in a current battery usage environment. This allows the user to know in advance when to replace the battery and to properly manage it.

Hereinafter, specific configurations and operations of components of the battery management system of FIG. 1 will be described.

When the numerical analysis result is verified as a laboratory result through electrical and chemical analysis of the battery, the controller 110 extracts the solution data of the battery and obtains a stored numerical simulation database.

The controller 110 counts the number of charges or discharges when the deviation between the reference data read from the verified numerical simulation database and the measured data read from the battery is within a preset range and the battery capacity satisfies the preset condition. Alternatively, an SOH prediction algorithm is performed during discharge.

Thereafter, the controller 110 performs a machine learning algorithm (Machine Learning Algorithm) for predicting the life state of the battery using a classifier according to the number of times of charge or discharge and the machine learning algorithm.

Meanwhile, when the deviation between the reference data and the measurement data is out of the preset range, the controller 110 emergencyly stops charging or discharging the battery through a safety algorithm.

The data input / output unit 120 receives data necessary for estimating the life state of the battery. In addition, the data input / output unit 120 may output data calculated in relation to the predicted life state of the battery or the battery.

The numerical simulation database 130 stores the solution data for the battery extracted through the electro-chemical analysis.

The error database 140 stores the error value calculated by the controller 110.

FIG. 5 shows a conceptual diagram of a lithium ion battery composed of an anode, a separator, and a cathode. Each electrode comprises an active material, a filler, a polymeric binder and an electrolyte. Graphite MCMB2528 and LiMn ₂ O ₄ are the main active materials for positive and negative electrodes. The electrolyte is a mixture of propylene carbonate (PC, 10% vol), ethylene carbonate (EC, 27% vol) and di-methyl carbonate (DMC, 63% vol). On the other hand, the porous separator is composed of a liquid electrolyte and p (VdF-HFP).

The physical phenomenon of the lithium ion battery may be represented by [Equation 1] to [Equation 7]. These represent physically dominant phenomena for voltage and mass transport (concentration) under electrolyte and solid state, current balancing inside batteries and energy balance based on thermodynamic theory.

a) Potential in the electrolyte

b) Potential in solid

c) Transport in the electrolyte

d) transport in solid

e) reaction rate

f) Current balance

g) energy balance

Where a is the surface area of the active material per electrode volume, c is the salt concentration of the electrolyte, cs is the lithium concentration at the solid insertion electrode, Cp is the heat capacity, D is the salt diffusion coefficient, Ds is the diffusion coefficient of lithium at the insertion electrode, f_ ± Average molar activity coefficient of the electrolyte, F is the Faraday constant, in is the current carrying perpendicular to the surface of the active material, i ₀ is the exchange current density, i ₂ is the current density (surface area) in the electrolyte, I is the total current density of the cell , j is the total flux by the reaction, n is the number of electrons involved in the reaction, Q is the heat generation rate, R is the general gas constant, s is the stoichiometric coefficient, positive for the anode reactant, and Time is the time. where t_i ^ 0 represents the number of transitions of species 'i' to solvent velocity, T is temperature and K, v is velocity, and zi is charge of ion i.

Α is the transfer coefficient, ε is the volume fraction (unless otherwise specified in the electrolyte), κ is the effective ion conductivity, ν is the number of moles of ions dissociated from the electrolyte, and σ is the effective electron of the porous electrode. Conductivity, Φ represents the potential. a represents an anode, c represents a cathode, i represents a species i, + represents an anode, and-represents a cathode.

The above equations are partial differential equations derived from the basic theory of lithium ion batteries. A commercial multiphysics package program is used to solve Equations 1 through 7, the governing partial differential equations of a battery cell taking into account internal electrochemical reactions. The computational reliability of the console in the current numerical model has been validated in previous reports by comparing the results with experiments on battery charge and discharge and surface temperature. Deviations between the experimental results and the simulation model were considered within tolerances for temperature and voltage.

6 is a diagram illustrating a lithium ion battery charging curve at various temperatures. 7 is a diagram illustrating a discharge curve of a lithium ion battery at various temperatures.

To obtain the predicted simulation data in a reliable numerical way, the numerical model of the battery pack must be optimally designed. It should have cells evenly balanced at surface temperature. Therefore, embodiments of the present invention may predict simulation data regarding SOC and SOH of internal performance of a battery cell at different environmental temperatures.

6 and 7 show changes in charge and discharge rates for different temperatures of a cylindrical lithium ion battery case. Each simulation assumes that the environmental temperature of the battery cell has remained constant from the beginning. Experimental results indicate that the lower the environmental temperature, the faster the charge and discharge rate. As the temperature decreases, the reaction rate also falls inside the battery, which is similar to the capacity reduction of older batteries.

8 shows the time (1 cycle) to reach 4V while charging at different temperatures. Fig. 9 shows the time (1 cycle) before reaching 3.5V while discharging at different temperatures.

8 and 9 show the time required to reach the target charge and discharge voltage at various environmental temperatures for quantitative analysis of the effect of temperature. The target voltage is determined at 4V and 3.6V, the top and bottom voltages of commonly used Li-ion batteries. Both figures show that the charge and discharge rates rise or fall almost linearly with different environmental temperatures. Formulas and values created by linear regression analysis are inserted into each figure. The temperature values and formulas of FIGS. 8 and 9 are derived over a temperature range between 10 degrees Celsius and 80 degrees Celsius.

As such, the four figures of FIGS. 6-9 show the predicted voltage change of the SOC and the charge or discharge time to reach the target voltage at different cell surface temperatures.

The method for predicting the life state of a battery according to an embodiment of the present invention is to provide a new algorithm for SOC and SOH prediction based on numerical sample cases with reliable big data predicted with respect to SOC and SOH.

On the other hand, there are two ways to predict the SOH of the battery cell at different temperatures. One method is to check the operating time or count the number of periodic operations to determine the remaining life of the battery cell under normal charging and discharging conditions. Another method is to provide an alarm or replacement signal to the BMS when abnormal operation continues for a certain time within tolerance. Embodiments of the present invention focus more on the second approach to derive SOH prediction algorithms based on normal data obtained through various reliable numerical analysis. SOC is therefore obtained by a numerical model validated over the normal temperature range. The validated numerical model also provides a sufficient database of commonly required voltages.

10 shows a numerical simulation flow diagram for obtaining a reliable reference database to be used for the SOH prediction algorithm of a battery. The SOH prediction algorithm should be coded in the controller 110 of the battery management system. The controller 110 detects an abnormal state that deviates from the normal database of voltages when the battery is charged and discharged at different temperatures.

In addition to the voltage change to be calculated, the database for the time to reach the required voltage at different temperatures provides a steady state for applying the SOH prediction algorithm. Therefore, the alarm notifies in advance of the abnormal state when the real-time measurement data in the battery management system 100 deviates from the proposed normal standard database. Repeated and accumulated data from Li-Ion battery packs can improve the database and error detection conditions through artificial intelligence (AI) algorithms, providing faster and more accurate SOH predictions. Data generated from existing data and artificial intelligence algorithms is repeatedly combined with new numerical simulation data. Through this, the newly generated data is classified, extracted, and manipulated to predict the SOH of the lithium ion battery cell.

As shown in FIG. 10, the battery management system 100 sets initial conditions (S101).

For example, the battery management system 100 may set boundary conditions and material properties.

The battery management system 100 performs a numerical analysis (S102). Specifically, the battery management system 100 performs an electro-chemical analysis. For example, electro-chemical analysis may be performed according to Species Transport Model, Electronic Potential Model, Chemical Reaction Model, Energy Balance Model.

The battery management system 100 checks whether it is integrated (S103).

If not, the battery management system 100 checks the mesh quality and the physical model (S104).

If the confirmation result (S103), if integrated, the battery management system 100 verifies the experimental results (Experimental Results) (S105).

The battery management system 100 confirms whether the reliability (S106).

As a result of the check (S106), if there is no reliability, the battery management system 100 performs a process S104 to check the mesh quality (Physical Quality) and the physical model (Physical Model).

If the check result (S106), if there is reliability, the solution data (Solution Data) is extracted (S107).

The battery management system 100 stores the extracted solution data in a verified numerical simulation database.

11 shows a flow diagram for an artificial intelligence (AI) algorithm based on verified numerical simulation data.

The battery management system 100 obtains training data including training data and numerical data that exist, in operation S201.

The battery management system 100 performs a machine learning process using learning data according to a machine learning algorithm (S201). For example, the machine learning algorithm may include a support vector machine, Bayes classifiers, artificial neural networks, decision trees, and the like.

Then, the battery management system 100 receives new data from the prediction algorithm of the battery life state (SOH) (S203). New data is received from each algorithm shown in FIGS. 12 to 16, respectively.

The classifier of the battery management system 100 classifies the received new data based on the learned machine learning algorithm (S204).

The battery management system 100 performs a process of predicting a life state of a battery (S205).

Meanwhile, the SOH prediction algorithm based on the simulated voltage will be described.

6 and 7 show that the characteristics of the lithium ion battery depend on the environmental temperature. Reliable databases calculated through numerical simulations can be used as standard reference data to determine the amount of deviation from the measured voltage at different temperatures. If the measured data do not follow the proposed voltage change at a certain temperature within tolerance, it may be determined that the operating battery cell may not be in a good life state.

Logical flow diagrams for predicting SOH are shown in FIGS. 12 and 13 for the case of charge and discharge. V _n is the reference data of the reference voltage accumulated by the reliable numerical simulation at a specific temperature, and V _r represents the measurement data for the real-time measured voltage of the battery cell. The deviation of the two values of V _n and V _r may be used to determine whether an alarm is sent only to the battery management system 100 or whether to stop the battery cell immediately. Only in the event of an alarm, the safety algorithm of the battery management system 100 should be selected and activated.

Since no battery cell can be stopped for any alarm, an algorithm for making the final decision is required, which is shown in FIG. 14. The algorithm for the final decision checks the accumulated error data with a certain number of cycles and if this value exceeds the maximum error value (ERRmax), an emergency stop should be executed.

The battery management system 100 receives a reference voltage (V _n ), a maximum allowable time (t _max ), a minimum allowable time (t _min ), a reference temperature (T _n ), and a reference current from the verified numerical simulation database 130. I _n ),… , Etc. are read (S301).

The battery starts charging (S302).

The battery management system 100 may measure the measured voltage V _r , the measured time t _r , the measured temperature T _r , the measured current I _r,. , And the like are read (S303).

The battery management system 100 calculates an _nth error value ERR _n according to Equation 8 below (S304). The battery management system 100 stores the calculated ERR _n in the error database 140.

Here, ERR _n is the n-th error value, V _n is the reference data of the reference voltage accumulated by the reliable numerical simulation at a specific temperature, and V _r represents the measurement data for the real-time measured voltage of the battery cell.

The battery management system 100 checks whether ERR _n is within a preset range (S305).

If the check result (S305), ERR _n is not within a predetermined range, the battery management system 100 checks whether the case where the ERR _n is not a predetermined range (S306).

If the check result (S306), if the case where ERR _n is not a predetermined range is a continuous event, the battery management system 100 performs an alarm (S307).

The battery management system 100 moves to the safety algorithm and performs the safety algorithm (S307). The safety algorithm is shown in FIG. 14.

If the check result (S305), the case where the ERR _n is not a predetermined range is not a continuous event, the battery management system 100 calculates a battery capacity (SOC) (S309).

Meanwhile, as a result of the check, if ERR _n is within a preset range, the battery management system 100 performs a process S309 for calculating a battery capacity SOC.

The battery management system 100 checks whether the calculated battery capacity is 100% (S310).

If the check result (S310), the calculated battery capacity is not 100%, the battery management system 100 is V _r , t _r , T _r , I _r ,. Go to S303 to read, etc. to perform.

As a result of the check, if the calculated battery capacity is 100%, the battery management system 100 counts the number of times of charging (S311).

The battery management system 100 stops charging (S312).

The battery management system 100 transmits all data to the machine learning algorithm (S313). The machine learning algorithm is shown in FIG.

The battery management system 100 moves to the SOH prediction algorithm upon discharge and performs the SOH prediction algorithm upon discharge (S314). The SOH prediction algorithm upon discharge is shown in FIG. 12.

The battery management system 100 reads V _n , t _max , t _min , T _n , I _n , SOC _min , and the like from the verified numerical simulation database 130 (S401).

The battery starts discharging (S402).

The battery management system 100 is configured such that V _r , t _r , T _r , I _r ,. , And the like are read (S403).

The battery management system 100 calculates ERR _n according to Equation 8 above (S404). The battery management system 100 stores the calculated ERR _n in the error database 140.

The battery management system 100 checks whether ERR _n is within a preset range (S405).

If the check result (S405), ERR _n is not within the predetermined range, the battery management system 100 checks whether the case where the ERR _n is not a predetermined range (S406).

If the check result (S406), if the case where the ERR _n is not a predetermined range is a continuous event, the battery management system 100 performs an alarm (S407).

The battery management system 100 moves to the safety algorithm and performs the safety algorithm. The safety algorithm is shown in FIG. 13 (S408).

If the check result (S406), the case where the ERR _n is not a predetermined range is not a continuous event, the battery management system 100 calculates a battery capacity (SOC) (S409).

Meanwhile, as a result of the check, if ERR _n is within a preset range, the battery management system 100 performs an operation S409 for calculating a battery capacity SOC.

The battery management system 100 checks whether the calculated battery capacity is equal to or less than SOC _min (S410).

The check result (S410). If the calculated battery capacity exceeds SOC _min , the battery management system 100 determines that V _r , t _r , T _r , I _r ,... Go to step S403 for reading, etc. to perform.

The check result (S410). If the calculated battery capacity is equal to or less than SOC _min , the battery management system 100 counts the number of times of discharge (S411).

The battery management system 100 stops discharging (S412).

The battery management system 100 transmits all data to the machine learning algorithm (S413). The machine learning algorithm is shown in FIG.

The battery management system 100 moves to the SOH prediction algorithm upon discharge and performs the SOH prediction algorithm upon discharge (S414). The SOH prediction algorithm upon discharge is shown in FIG. 12.

The battery management system 100 reads the n th ERR _n and the n-1 th ERR _n-1 from the error database (S501).

The battery management system 100 calculates the sum DEV _n of the deviations of the _nth ERR _n and the n−1th ERR _n ₋₁ according to Equation 9 below (S502).

The battery management system 100 checks whether the maximum error ERR _max is less than or equal to the sum of DEV _n from the 1 st to k th (S503).

When the maximum error ERR _max exceeds the sum of DEV _n from 1 to kth, the battery management system 100 determines the battery capacity (at the time of charge and discharge shown in FIGS. 11 and 12). SOC) and moves to the calculation block (S504).

As a result of the confirmation (S503), if the maximum error ERR _max is less than or equal to the sum of DEV _n from 1st to kth, the battery management system 100 transmits all data to the machine learning algorithm shown in FIG. 10 (S505).

The battery management system 100 performs an emergency stop (S506).

Meanwhile, the SOH prediction algorithm according to the charge and discharge time will be described. 11 and 12 illustrate an SOH prediction algorithm in which SOH can be predicted based on voltage data having a deviation between measured data and reference data. In addition to the voltage data for evaluating SOH, there is time to measure the charge and discharge rates for aged and malfunctioning battery cells. As a battery cell reaches its end of life or reaches its end of life, the charging time or the time for charging to an upper or lower voltage level is shortened.

15 and 16 show logic flow diagrams for predicting SOH when charged and discharged based on time. t _max and t _min are the maximum and minimum allowable times to reach the charged and discharged target voltage. Values exceeding these maximum and minimum allowable target values mean that the state of the battery cell is abnormal and an algorithm for controlling this limit must also be implemented.

The battery management system 100 is configured from the verified numerical simulation database V _n , t _max , t _min , T _n , I _n ,. , Etc. are read (S601).

The battery management system 100 starts charging the battery (S602).

The battery management system 100 is configured to calculate V _r , t _r , T _r , I _r ,. , And the like are read (S603).

The battery management system 100 checks whether V _r is less than or equal to V _n (S604).

If V _r is equal to or less than V _n , the battery management system 100 checks whether t is equal to or less than t _max (S605).

If t is less than or equal to t _max , the battery management system 100 determines V _r , t _r , T _r , I _r ,... The process proceeds to step S603 of reading, etc.

On the other hand, if the check result (S604), V _r is not V _n or less, the battery management system 100 checks whether t is less than t _min (S606).

If the check result (S606), t is less than t _min , the battery management system 100 checks whether the event is a continuous event (S607).

On the other hand, if the check result (S605), t is not less than t _max , the battery management system 100 performs a process S607 to determine whether the event is a continuous event.

If the check result (S607), the event is not a continuous event, the battery management system 100 is V _r , t _r , T _r , I _r ,. The process proceeds to step S603 of reading, etc.

On the other hand, if the check result (S607), a continuous event, the battery management system 100 performs an alarm (S608).

The battery management system 100 moves to the safety algorithm and performs the safety algorithm (S609). The safety algorithm is shown in FIG. 13.

If the check result (S606), t is not less than t _min , the battery management system 100 calculates a battery capacity (SOC) (S610).

The battery management system 100 checks whether the calculated battery capacity is 100% (S611).

As a result of the check (S611), if the calculated battery capacity is not 100%, the battery management system 100 in the battery V _r , t _r , T _r , I _r ,. The process proceeds to step S603 of reading, etc.

As a result of the check (S611), if the calculated battery capacity is 100%, the battery management system 100 stops charging (S612).

The battery management system 100 counts the number of charges (S613).

The battery management system 100 transmits all data to the machine learning algorithm (S614). The machine learning algorithm is shown in FIG.

The battery management system 100 moves to the SOH prediction algorithm upon discharge and performs the SOH prediction algorithm upon discharge (S615). The SOH prediction algorithm upon discharge is shown in FIG. 13.

The battery management system 100 is configured from the verified numerical simulation database V _n , t _max , t _min , T _n , I _n , SOC _min ,. , Etc. are read (S701).

The battery management system 100 starts discharging the battery (S702).

The battery management system 100 is configured such that V _r , t _r , T _r , I _r ,. , And the like are read (S703).

The battery management system 100 checks whether V _n is less than or equal to V _r (S704).

If V _n is equal to or less than V _r , the battery management system 100 checks whether t is equal to or less than t _max (S705).

If t is less than or equal to t _max , the battery management system 100 determines V _r , t _r , T _r , I _r ,... Go to step S703 to read, and so on.

On the other hand, if the check result (S704), V _n is not less than V _r , the battery management system 100 checks whether t is less than t _min (S706).

If the check result (S706), t is less than t _min , the battery management system 100 checks whether a continuous event (S707).

On the other hand, if the check result (S705), t is less than or equal to t _max , the battery management system 100 performs a process S707 to determine whether the event is continuous.

If the check result (S707), the event is not continuous, the battery management system 100 is V _r , t _r , T _r , I _r ,. Go to step S703 to read, and so on.

On the other hand, if the check result (S707), a continuous event, the battery management system 100 performs an alarm (S708).

The battery management system 100 moves to the safety algorithm and performs the safety algorithm (S709). The safety algorithm is shown in FIG. 13.

If the check result (S706), t is not less than t _min , the battery management system 100 calculates a battery capacity (SOC) (S710).

The battery management system 100 checks whether the calculated battery capacity is equal to or less than SOC _min (S711).

As a result of the check (S711), if the calculated battery capacity is not equal to or less than SOC _min , the battery management system 100 determines that V _r , t _r , T _r , I _r ,... Go to step S703 to read, and so on.

If the check result (S711), the calculated battery capacity is less than SOC _min , the battery management system 100 stops the discharge (S712).

The battery management system 100 counts the number of times of discharge (S713).

The battery management system 100 transmits all data to the machine learning algorithm (S714). The machine learning algorithm is shown in FIG.

The battery management system 100 moves to the SOH prediction algorithm when charging and performs the SOH prediction algorithm when charging (S615). The SOH prediction algorithm on charge is shown in FIG. 13.

As described above, the battery life state prediction method according to an embodiment of the present invention relates to a battery management system (BMS) algorithm for predicting the life state (SOH) of a lithium ion battery based on numerical analysis data. SOH prediction method according to an embodiment of the present invention was verified the reliability compared to the experimental experiment. A reliable numerical simulation database calculated through numerical simulation can be used as standard reference data to confirm the amount of deviation from measured voltages at different temperatures. If the measured data do not follow the voltage change over a particular temperature, it may be determined that the working battery cell may not be in a good life state.

In addition to voltage data for evaluating SOH, there is time to measure charge and discharge rate rates for aged and malfunctioning battery cells. An embodiment of the present invention may provide a logic algorithm that predicts SOH when charged and discharged over time.

The SOH prediction method uses the voltage change and the charge and discharge time to predict the life of the battery. These two SOH prediction methods indicate that algorithms based on numerical big data can be applied to other Li-ion battery cells to predict SOC and SOH. The error value ERR in the SOH prediction method may be accumulated by itself using an improved artificial intelligence (AI) algorithm, such as deep learning, to be more appropriately improved.

Although embodiments of the present invention have been described with reference to the above contents, those skilled in the art will appreciate that the present invention can be implemented in other specific forms without changing the technical spirit or essential features.

Therefore, the embodiments described above are to be understood as illustrative and not restrictive in all respects, and the scope of the present invention described in the above detailed description is represented by the following claims, and the meanings of the claims and All changes or modifications derived from the scope and equivalent concepts thereof should be construed as being included in the scope of the present invention.

Claims

In the method of predicting the life state of a battery performed by a battery management system,

If the numerical analysis result is verified as a laboratory result through electrical and chemical analysis of the battery, extracting the solution data of the battery to obtain a stored numerical simulation database;

Counting the number of charges or discharges when the deviation between the reference data read from the verified numerical simulation database and the measurement data read from the battery is within a preset range and the battery capacity satisfies a preset condition; And

Predicting a life state of a battery by using a classifier according to the number of charges or discharges and a previously learned machine learning algorithm;

Including,

The obtaining step,

Setting an initial condition for the battery;

Performing numerical analysis through electrical and chemical analysis on the battery;

Verifying the performed numerical analysis result as an experimental result;

Extracting solution data of a battery for the numerical analysis result when the verification is completed; And

Storing the extracted solution data in a verified numerical simulation database and obtaining from the verified numerical simulation database;

Life state prediction method of the battery based on the numerical simulation data comprising a.
The method of claim 1,

Performing the numerical analysis,

Numerical simulation that performs numerical analysis using at least one of Species Transport Model, Electronic Potential Model, Chemical Reaction Model, and Energy Balance Model How to predict the life state of a data-based battery.
In the method of predicting the life state of a battery performed by a battery management system,

If the numerical analysis result is verified as a laboratory result through electrical and chemical analysis of the battery, extracting the solution data of the battery to obtain a stored numerical simulation database;

Counting the number of charges or discharges when the deviation between the reference data read from the verified numerical simulation database and the measurement data read from the battery is within a preset range and the battery capacity satisfies a preset condition; And

Predicting a life state of a battery by using a classifier according to the number of charges or discharges and a previously learned machine learning algorithm;

Including,

The counting step,

Numerical simulation data-based battery that counts the number of charges and stops charging if the battery is charging and the battery capacity is fully charged, and counts the number of discharges and stops discharging if the battery is discharged and the battery capacity is the minimum battery capacity. Method for predicting lifespan status.
In the method of predicting the life state of a battery performed by a battery management system,

If the numerical analysis result is verified as a laboratory result through electrical and chemical analysis of the battery, extracting the solution data of the battery to obtain a stored numerical simulation database;

Counting the number of charges or discharges when the deviation between the reference data read from the verified numerical simulation database and the measurement data read from the battery is within a preset range and the battery capacity satisfies a preset condition; And

Predicting a life state of a battery by using a classifier according to the number of charges or discharges and a previously learned machine learning algorithm;

Including,

The counting step,

Reading reference data from a validated numerical simulation database;

Reading measurement data from a battery that is being charged or discharged;

Calculating a deviation between the read reference data and the read measurement data;

Checking whether the calculated deviation is within a preset range and calculating a battery capacity if the calculated deviation is within a preset range; And

Counting the number of charges or the number of discharges and stopping charging or discharging when the calculated battery capacity is full or less than the minimum battery capacity;

Life state prediction method of the battery based on the numerical simulation data comprising a.
The method of claim 1,

The predicting step,

Numerical estimate of battery life using any machine learning algorithm from Support Vector Machine, Bayes Classifiers, Artificial Neural Networks, or Decision Tree. How to predict the life status of batteries based on simulation data.
In the method of predicting the life state of a battery performed by a battery management system,

If the numerical analysis result is verified as a laboratory result through electrical and chemical analysis of the battery, extracting the solution data of the battery to obtain a stored numerical simulation database;

Counting the number of charges or discharges when the deviation between the reference data read from the verified numerical simulation database and the measurement data read from the battery is within a preset range and the battery capacity satisfies a preset condition; And

Predicting a life state of a battery by using a classifier according to the number of charges or discharges and a previously learned machine learning algorithm;

Including,

Emergency stop of charging or discharging of the battery through a safety algorithm when the deviation between the reference data and the measurement data is out of a predetermined range;

Life time prediction method of the battery based on the numerical simulation data further comprising.
The method of claim 6,

The emergency stop step,

Reading an nth error value and an n-1th error value from an error database;

calculating a sum of error deviation values of deviations of the n th error value and the n-1 th error value;

Checking whether a preset maximum error value is equal to or less than the sum of the calculated error deviation values;

Calculating a battery capacity at the time of charging or discharging, when the preset maximum error value is not equal to or less than the sum of the calculated error deviation values; And

Emergencyly stopping the charging or discharging operation of the battery when a predetermined maximum error value is less than or equal to the calculated error value sum;

Life state prediction method of the battery based on the numerical simulation data comprising a.
In the method of predicting the life state of a battery performed by a battery management system,

If the numerical analysis result is verified as a laboratory result through electrical and chemical analysis of the battery, extracting the solution data of the battery to obtain a stored numerical simulation database;

Compare the measured data read from the battery with the reference data read from the verified numerical simulation database, and the charging or discharging time is within the maximum and minimum allowable time for reaching the target voltage, and the battery capacity satisfies the preset conditions Counting the number of charges or discharges; And

Predicting a life state of a battery by using a classifier according to the counted number of charges or discharges and a previously learned machine learning algorithm;

Including,

The obtaining step,

Setting an initial condition for the battery;

Performing numerical analysis through electrical and chemical analysis on the battery;

Verifying the performed numerical analysis result as an experimental result;

Extracting solution data of a battery for the numerical analysis result when the verification is completed; And

Storing the extracted solution data in a verified numerical simulation database and obtaining from the verified numerical simulation database;

Life state prediction method of the battery based on the numerical simulation data comprising a.
The method of claim 8,

Performing the numerical analysis,

Numerical simulation that performs numerical analysis using at least one of Species Transport Model, Electronic Potential Model, Chemical Reaction Model, and Energy Balance Model How to predict the life status of data-based batteries.
In the method of predicting the life state of a battery performed by a battery management system,

If the numerical analysis result is verified as a laboratory result through electrical and chemical analysis of the battery, extracting the solution data of the battery to obtain a stored numerical simulation database;

Compare the measured data read from the battery with the reference data read from the verified numerical simulation database, and the charging or discharging time is within the maximum and minimum allowable time for reaching the target voltage, and the battery capacity satisfies the preset conditions Counting the number of charges or discharges; And

Predicting a life state of a battery by using a classifier according to the counted number of charges or discharges and a previously learned machine learning algorithm;

Including,

The counting step,

Numerical simulation data-based battery that counts the number of charges and stops charging if the battery is charging and the battery capacity is fully charged, and counts the number of discharges and stops discharging if the battery is discharged and the battery capacity is the minimum battery capacity. Method for predicting lifespan status.
In the method of predicting the life state of a battery performed by a battery management system,

If the numerical analysis result is verified as a laboratory result through electrical and chemical analysis of the battery, extracting the solution data of the battery to obtain a stored numerical simulation database;

Compare the measured data read from the battery with the reference data read from the verified numerical simulation database, and the charging or discharging time is within the maximum and minimum allowable time for reaching the target voltage, and the battery capacity satisfies the preset conditions Counting the number of charges or discharges; And

Predicting a life state of a battery by using a classifier according to the counted number of charges or discharges and a previously learned machine learning algorithm;

Including,

The counting step,

Reading reference data from a validated numerical simulation database;

Reading measurement data from a battery that is being charged or discharged;

Comparing measured data read from a battery with reference data read from the verified numerical simulation database;

Check whether the charge or discharge time is within the maximum and minimum allowable time to reach the target voltage, and if the charge or discharge time is within the maximum and minimum allowable time to reach the target voltage and the battery capacity meets the preset conditions Calculating a dose; And

Counting the number of charges or the number of discharges and stopping charging or discharging when the calculated battery capacity is less than or equal to full charge or minimum battery capacity;

Life state prediction method of the battery based on the numerical simulation data comprising a.
The method of claim 8,

The predicting step,

Numerical estimate of battery life using any machine learning algorithm from Support Vector Machine, Bayes Classifiers, Artificial Neural Networks, or Decision Tree. How to predict the life status of batteries based on simulation data.
In the method of predicting the life state of a battery performed by a battery management system,

If the numerical analysis result is verified as a laboratory result through electrical and chemical analysis of the battery, extracting the solution data of the battery to obtain a stored numerical simulation database;

Compare the measured data read from the battery with the reference data read from the verified numerical simulation database, and the charging or discharging time is within the maximum and minimum allowable time for reaching the target voltage, and the battery capacity satisfies the preset conditions Counting the number of charges or discharges; And

Predicting a life state of a battery by using a classifier according to the counted number of charges or discharges and a previously learned machine learning algorithm;

Including,

Emergency stop of charging or discharging of the battery via a safety algorithm if the charge time or discharge time deviates from the maximum and minimum allowable time for reaching the target voltage;

Life time prediction method of the battery based on the numerical simulation data further comprising.
The method of claim 13,

The emergency stop step,

Reading an nth error value and an n-1th error value from an error database;

calculating a sum of error deviation values of deviations of the n th error value and the n-1 th error value;

Checking whether a preset maximum error value is equal to or less than the sum of the calculated error deviation values;

Calculating a battery capacity at the time of charging or discharging, when the preset maximum error value is not equal to or less than the sum of the calculated error deviation values; And

Emergencyly stopping the charging or discharging operation of the battery when a predetermined maximum error value is less than or equal to the calculated error value sum;

Life state prediction method of the battery based on the numerical simulation data comprising a.